Civil Justice Reform - Law Firms The Duty To Preserve Evidence In A Digital World

In a series of widely publicized cases, courts have issued harsh sanctions for the destruction of electronic evidence. As companies wrestle with the uncertain scope of the duty to preserve in the context of burgeoning digital data, proposed amendments to the Federal Rules of Civil Procedure promise clearer guidance. Unfortunately, the amendments largely relegate parties to the underdeveloped body of common law on the subject.

The Scope Of The Duty To Preserve

The scope of a party's duty to preserve evidence remains unsettled, but many courts have articulated a test whereby a party must preserve all evidence the party knows or has reason to know is (i) relevant to a pending action, (ii) reasonably calculated to lead to the discovery of admissible evidence, (iii) reasonably likely to be requested in discovery, and/or (iv) the subject of a pending discovery request. See, e.g., Zubulake v. UBS Warburg LLC, No. 02 Civ. 1243, 2003 WL 22410619, *3 (S.D.N.Y. Oct. 22, 2003); Danis v. USN Communications, Inc., No. 98 C 7482, 2000 WL 1694325, *32 (N.D. Ill. Oct. 23, 2000). However, courts have done little to flesh out the meaning of each prong of this test. Take, for example, the third prong. As most litigators know all too well, it is often reasonably likely that an adversary will request items that are unlikely to lead to the discovery of admissible evidence and, hence, not properly discoverable. The test articulated by these courts may ensure the preservation of such items until a court can resolve their discoverability. However, it is unrealistic to expect a party to anticipate everything reasonably likely to be requested in discovery.

Deleted Data, Archived Data, And Meta Data

Moreover, the manner in which computers create and store electronic information will frequently make it infeasible to preserve all evidence in this potentially limitless category. For example, when a witness creates an electronic document (whether it be an e-mail message, a word processing document, or a database), that document will contain a variety of "meta data" or information about the document that was never meant to be seen by the external world. Meta data includes such information as file names, directory paths, names of persons who created or updated a document, and dates of creation and update. This type of information is constantly evolving as documents are accessed and used. Furthermore, when a witness deletes a document or a draft of a document, the deleted data continues to reside on the computer's hard drive. However, as new information is saved and randomly allocated to available storage space on the hard drive, a deleted document may be overwritten partially or entirely. In addition, a document that has been deleted from a hard drive may also exist in archival media, including on tapes containing periodic back-ups. Such back-up tapes are often recycled for reuse, which may render previously stored information irretrievable. Regardless of whether deleted data has been overwritten, the recovery of deleted data from a hard drive or from back-up tapes may be costly and disruptive.

In most litigation, one would expect the discoverable evidence to consist of data created within that part of a document meant to be viewed by the external world, and one would expect discoverable documents to continue to exist in a reasonably accessible form in the possession of one or more witnesses. Accordingly, most litigation should not require production of meta data, deleted data, or archived data. Nevertheless, a number of lawyers are routinely requesting these items. Furthermore, even where these items are not explicitly requested, a discovery target will often be confronted with the difficult task of determining whether a particular case is one that warrants searching for such items.

Accessibility And Preservation

In addressing these problems, the central advance in the law of discovery is the recognition that a party's discovery obligations should vary depending upon the degree to which data is deemed "accessible" or "inaccessible." This distinction was articulated in the highly influential Zubulake case, in which the court held that whether discovery of electronic data should be deemed unduly burdensome turns primarily on whether such data is "accessible" or "inaccessible." The court deemed electronic data to be typically accessible due to the capabilities of search engines but also noted that data contained in certain back-up tapes and erased, fragmented, or damaged data may be sufficiently more expensive and cumbersome to restore to a usable format such that they should be deemed inaccessible. See Zubulake v. UBS Warburg LLC, 217 F.R.D. 309, 318-20 (S.D.N.Y. 2003).

Although this classification helps in determining whether data should be discoverable absent cost-shifting to the discovering party, it does not provide as ready a guide for a party attempting to determine whether data that might be deemed inaccessible should be preserved. For example, in addressing the scope of a party's discovery obligations, one court recognized the inaccessible nature of data residing on back-up tapes by likening the recovery of such data to "dumpster diving" but, nevertheless, ordered such discovery subject to cost-shifting. See Rowe Entmt., Inc. v. William Morris Agency, Inc., 205 F.R.D. 421, 431 (S.D.N.Y. 2002). As noted earlier, courts articulating the scope of the duty to preserve go one step further by indicating that the duty to preserve applies to evidence that is reasonably likely to be requested in discovery.

The Zubulake court addressed the duty to preserve by outlining one of presumably multiple ways to discharge the duty. The course of action outlined by the Zubulake court consists of preserving all back-up tapes in existence to the extent such tapes can be identified as containing information for "relevant personnel" and making a mirror image of the hard drives for the entire company's computer system. See Zubulake v. UBS Warburg LLC, 220 F.R.D. 212, 218 (S.D.N.Y. 2003). However, because a mirror image of a hard drive captures every bit of information (active data, meta data, and deleted data) residing on the hard drive, taking mirror images of the entire computer system may come close to preserving every document in the company. Yet, courts seemingly agree that a party does not have a duty to preserve every piece of paper. See, e.g., Wiginton v. Ellis, No. 02 C 6832, 2003 WL 22439865, *4 (N.D. Ill. Oct. 27, 2003); Shamis v. Ambassador Factor Corp., 34 F. Supp. 2d 879, 889 (S.D.N.Y. 1999). One court has suggested that if the type of approach suggested by Zubulake is cost prohibitive, a party may discharge its duty to preserve by performing searches for relevant information to capture and preserve information that is likely discoverable before it is lost in routine operations. See Wiginton, 2003 WL 22439865, at *5.

In this unsettled environment, the Advisory Committee on the Federal Rules of Civil Procedure has proposed amendments specifically addressing electronic discovery. One such amendment to Rule 26 adopts the accessible/inaccessible distinction by providing that a party need not produce discovery of electronically stored information from sources the party identifies as not reasonably accessible because of undue burden or cost. Under this proposal, such an identification does not render such information absolutely immune from discovery. Rather, if either party files a motion, the discovery target will bear the burden of proving that the information is not reasonably accessible because of undue burden or cost, and, even if that showing is made, the court may order that the discovery be produced if good cause is shown. Proposed Fed. R. Civ. P. 26(b)(2).

Unfortunately, even under this amendment, the accessible/inaccessible distinction will be of limited utility in determining the scope of a party's duty to preserve. Indeed, the Advisory Committee Note to Amended Rule 26(b)(2) expressly states that identifying sources of electronically stored information as inaccessible will not eliminate any common law or statutory duties to preserve such information, and that the facts of the particular case will determine whether a party must preserve unsearched sources of potentially responsive information believed to be not reasonably accessible. Proposed Fed. R. Civ. P. 26(b)(2), advisory committee note.

Safe Harbor For Routine Operations

Nevertheless, included within the proposed amendments is a safe harbor for the destruction of certain data according to routine, good-faith operation of an electronic system. Specifically, an Amended Rule 37(f) provides: "Absent exceptional circumstances, a court may not impose sanctions under these rules on a party for failing to provide electronically stored information lost as a result of the routine, good-faith operation of an electronic information system." Proposed Fed. R. Civ. P. 37(f).

This proposal also suffers from problems that will limit its utility. One fundamental problem with the proposal is that it limits sanctions that may be imposed under the rules of civil procedure only, whereas the rules of civil procedure are not the source of the duty to preserve evidence or of sanctions enforcing that duty (absent violation of an order compelling production). Indeed, the Advisory Committee Note expressly states that the proposed rule would not affect other sources of authority to impose sanctions. Proposed Fed. R. Civ. P. 37(f), advisory committee note. Thus, the proposed rule appears merely to offer guidance rather than instruction.

Furthermore, in attempting to apply this safe harbor to meta data, deleted data, or data residing on back-up tapes that are the most obvious candidates for information that would be lost through "routine É operation of an electronic system," a party must remember that its operation of the system must be in "good faith" to qualify for the safe harbor and that, even then, the party may lose the protection of the safe harbor in "exceptional circumstances." Proposed Fed. R. Civ. P. 37(f). Thus, at most, the safe harbor would appear to establish a presumption that loss of information through routine operations should generally not be grounds for sanctions. However, the proposed rule provides no indication of when a party would have to suspend routine operations to discharge its duty to preserve. The Advisory Committee Note merely states that whether a party must take steps to prevent the loss of information from sources not believed to be reasonably accessible will depend upon the facts of each case and hints that one factor influencing this determination should be whether that information is likely to be discoverable and unavailable from other sources. Proposed Fed. R. Civ. P. 37(f), advisory committee note.


Accordingly, companies attempting to comply with the duty to preserve must determine what types of cases are likely to implicate meta data, deleted data, and data residing on back-up tapes. This determination is particularly difficult given that the use of these types of data in litigation is a relatively new phenomenon. Nevertheless, a party may wish to consider whether there are questions about the authenticity or origins of or motives behind a document, the individual storage and deletion practices of the key witnesses as a general matter, and whether witnesses are likely to have deleted discoverable information in the particular case at hand. In addition, to the extent a particular case is deemed reasonably likely to implicate these types of data, a party will need to determine what efforts it must devote to locating and preserving such data before it is lost in routine operation. Here, parties must choose between the approaches outlined in Zubulake and Wiginton, as well as other potential options. Finally, a party attempting to discharge its duty to preserve must not lose sight of the need to locate and take appropriate steps to preserve all accessible data, which may be widely distributed in a vast organization.

Published December 1, 2005.